Robotics Policy Evaluation Engineer

Walden Robotics; San Francisco (CA); On-site; USD 140,000 - 190,000; Full time

Job summary

Walden Robotics is seeking an engineer to build tooling for the autonomy team to measure whether our robot policies are actually getting better—in simulation and on real hardware. This is a high-leverage engineering role at the center of the research loop, with mentorship and room to grow quickly alongside senior engineers.

You will own end-to-end development of evaluation pipelines, dashboards, and checks, enabling researchers to quantify improvements and compare experiments reliably.

Qualifications

• Software fundamentals for reliable tooling and engineering practices.
• Hands-on experience with robot policies or robotic systems evaluating or debugging them.
• Genuine interest in ML systems and evaluation in real robots.
• Ability to take a well-scoped problem and deliver end to end.
• Eagerness to learn quickly from feedback and code review.

Responsibilities

• Policy Evaluation: Build and maintain pipelines that evaluate robot policies in sim and on hardware.
• Metrics for Robot Behavior: Implement task-success metrics, dashboards, and automated checks.
• Iteration Speed: Improve the experiment loop speed and reliability for faster evaluation.
• Debugging: Identify issues across the evaluation stack and policy rollouts.
• Growth: Learn the autonomy stack and grow ownership alongside senior engineers.

Skills

Software fundamentals

Robot policies

ML systems

Delivery

Curiosity & drive

Job description

Walden Robotics is seeking an engineer to build tooling for the autonomy team to measure whether our robot policies are actually getting better—in simulation and on real hardware. This is a high-leverage engineering role at the center of the research loop, with mentorship and room to grow quickly alongside senior engineers.

You will own end-to-end development of evaluation pipelines, dashboards, and checks, enabling researchers to quantify improvements and compare experiments reliably.